ResearchPolicy42 (2013) 1496–1510
ContentslistsavailableatSciVerseScienceDirect
Research
Policy
jou rn a l h o m e pa g e :w w w . e l s e v i e r . c o m / l o c a t e / r e s p o l
Innovation
and
firm
value:
An
investigation
of
the
changing
role
of
patents,
1985–2007
Sharon
Belenzon
a,1,
Andrea
Patacconi
b,∗aFuquaSchoolofBusiness,DukeUniversity,USA
bUniversityofAberdeenBusinessSchool,UK
a
r
t
i
c
l
e
i
n
f
o
Articlehistory:
Received17March2010
Receivedinrevisedform29April2013 Accepted3May2013
Available online 16 June 2013
JELclassification: O31 O32 O16 Keywords: Firmvalue
Patentportfoliovaluation Patentpolicy
a
b
s
t
r
a
c
t
Thispaperexamineshowtherelationshipbetweenfirmvalueandpatent-basedindicatorsofinventive activityhaschangedovertime.Weusedatafrommorethan33,000mergersandacquisitionsdeals between1985and2007,anddistinguishbetweenAmerican(USPTO)andEuropean(EPO)patents.Our resultsindicatethatovertimeEPOpatentshavebecomethedominantindicatorofinnovativeactivity, whileUSPTOpatentshavenoeffectonfirmvalueneartheendofthesampleperiod.Theresultsarerobust tocontrollingforcitationsandareespeciallystrongforsmallfirms,forfirmsoperatinginthedrugand chemicalindustries,andwhentargetandacquiringfirmsoperateindifferentindustriesorcountries.
© 2013 Elsevier B.V. All rights reserved.
1. Introduction
In evaluating potentialtargets for mergers and acquisitions
(M&A),acquiringfirmsmusttakeseveralfactorsintoaccount:the
valueofthetargetcompany’stangibleandintangibleassets,the
potentialforsynergiesbetweenthemergingfirms,cost-of-capital
andtaxconsiderations,andmore.Afactorthathasbecome
increas-inglyimportantin recent years isintellectual property.A large
patentportfoliocanpositivelyinfluenceafirm’svaluationinM&A
foranumber ofreasons.Patentsarelegaltitlesthatgrant their
ownersatemporarymonopolypowerovertheuseofaninvention.
Thus,theycanbeusedtoenforceexclusivity,toextract
licens-ingincome,orasabargainingchipinnegotiations.Wereferto
theseusesofintellectualpropertyasthemonopolyorlegalvalueof
patents.Patentscanalsobecorrelatedwithfirmvaluebecausethey
provideinformation aboutthequalityof inventionsand overall
“corporatetechnologicalstrength”(Long,2002;Narinetal.,1987).
Patentingisexpensive,thusonlyinventionsthataredeemed
suf-ficientlyvaluablearepatented.Moreover,examinersmustcheck
∗Correspondingauthorat: UniversityofAberdeenBusinessSchool,Edward
WrightBuilding,DunbarStreet,AberdeenAB243QY,UK.Tel.:+4401224273423.
E-mailaddresses:sharon.belenzon@duke.edu(S.Belenzon),
a.patacconi@abdn.ac.uk(A.Patacconi).
1 Address:DukeUniversity:TheFuquaSchoolofBusiness,100FuquaDrive, Durham,NC27708,USA.Tel.:+19196607845.
thatinventionssatisfyminimumstandardsof noveltyand
non-obviousness.More broadly,largepatent portfolios maysuggest
thatfirmsarewell-managedandpossessvaluableR&Dcapabilities.
Thus,largepatentportfolioscansignaltechnologicalstrength.
However,recentpatentandadministrativereformshavemost
likely affected both the monopoly and the informational (or
“quality-signaling”) value of patents. There is a broad
consen-susamongacademicsandpractitionersthatpatentprotectionin
mostOECDcountrieshasbeenconsiderablystrengthenedsincethe
1980s(JaffeandLerner,2004;Lerner,2002).Importantstepsinthis
directionhavebeenthecreationin1982ofa“pro-patent”Courtof
AppealsintheU.S.,thesubsequentupwardharmonizationof
intel-lectualpropertyprotectionstandardsviainternationalagreements,
andtheextensionofpatentabilitytonewsubjectmattersuchas
biotechnologyandsoftware(intheU.S.,alsobusinessmethods).2
Thus,themonopolyvalueofpatentshasmostlikelyincreasedin
thelastthreedecades.
At the same time, patent examination standards appear to
havedeclined.JaffeandLerner(2004)provideseveralexamples
ofobviousornon-novelideasthatweregrantedpatentprotection,
includinga“MethodtoSwingonaSwing”,the“SealedCrustless
Sandwich”,Rambus’sdivisionalapplicationsondevelopmentsof
2Recentjudicialdecisions,however,havearguablyrestrictedtherangeof busi-nessmethodsthatarepatentableintheU.S.
0048-7333/$–seefrontmatter© 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.respol.2013.05.001
itsDRAMtechnology,and,perhapsmorecontroversially,one-click
businessmodels. Importantly,thedecline in examination
stan-dardsappearstohavebeenpronouncedespeciallyintheU.S.,due
toanumberofinstitutional,legalandculturalfactors(Guellecand
vanPottelsberghedelaPotterie,2007;JaffeandLerner,2004).Thus,
patent“quality”(broadly intendedtoencompassboth the
tech-nologicalandvaluedimensionsofpatentedinventions)mayhave
declinedmoreintheU.S.thaninothermajorjurisdictions.
Motivatedbytheseconsiderations,thepresentpaperexamines
howtherelationshipbetweenfirmvalueandpatent-based
indi-catorsofinventiveactivityhaschangedinthelastthreedecades.
We use datafrom more than 33,000 mergers and acquisitions
dealsbetween1985and2007,anddistinguishbetweenAmerican
(USPTO)andEuropean(EPO)patents.Wefindthatwhiletheeffect
ofEPOpatentsonacquisitionvaluehasremainedessentiallystable
overthistimeperiod,theeffectofUSPTOpatentshasdeclined
sub-stantiallyandisessentiallyzeroneartheendofthesampleperiod.3
Giventhatpatentprotectionrightswerestrengthenedduringthis
timeperiod,ourresultssuggestthatanincreasingnumberofUSPTO
patentsweregrantedforinventionsthathadlittleornoeconomic
ortechnologicalvalue.USPTOpatentsseemtobelosingtheirrole
assignalsoffirms’inventiveeffortsandvalue.
Therobustnessofourfindingsisprobedinseveralways.
Qual-itativelysimilarresultsare obtainedwhen patentsareweighed
bycitations, although citation-weighinggenerally increasesthe
importance of USPTO patents. This is expected since citations
proxyforpatentqualityandpatentqualitymaytosomeextent
beobservedby theacquiringfirm (Hallet al., 2005).Our
find-ingsaredrivenbypatentsownedbysmallfirms.Thisisalsonot
surprisinggiventhatintellectualpropertyandsignaling
consid-erationsarearguablymostimportantinthestart-upphaseof a
firm’slife.Thereisalsosubstantialheterogeneityacrossindustries.
ThedeclineintheimportanceofUSPTOpatentsismostevidentin
drugsandchemicals,whilenosuchdeclineisfoundincomputers
andelectronics.4Wefindsimilarresultsinbiotechnologyandmore
traditionalsegmentsofthedrugandchemicalindustries.Thisdoes
notsupportthepresumptionthatU.S.patentqualitydeclinedmost
sharplyinnewtechnologyfields(Hall,2007;Merrilletal.,2004).
Finally,EPOpatentsappeartohavebecomemoreimportantover
time,relativetoUSPTOpatents,whenacquiringandtargetfirms
belongtodifferentindustriesoroperateindifferentcountries.We
conjecturethat,inthesedeals,becauseacquiringfirmsmayfindit
moredifficulttoevaluateatargetfirm’stechnology,EPOpatents
maybeperceivedas“safer.”TheEPOsystemmaybeprovidingsome
qualitycertification–acertificationwhichbecomesmore
impor-tantasuncertaintyandinformationalasymmetriesgrow(Akerlof,
1970).
Theremainderofthepaperisorganizedasfollows.Thenext
sectiondiscusses somefeatures ofthe Americanand European
patentsystemsthatmotivateourempiricalanalysis.Sections3and
4describethedataandeconometricspecification.Section5reports
theresults.Section6concludes.
2. Backgroundandpredictions
2.1. Thechanginginstitutionalenvironment
Thereiswidespreadconsensusamongacademicsand
practi-tioners that, starting from the early1980s, various patent and
3 However,wealsofindalargeincreaseinUSPTOpatentvaluationsinthelate 1990s.Thisincreasewasmostlikelyaconsequenceoftheexuberancethat charac-terizedofthedot-comyears.
4 Insoftware,atechnologyareawherepatentinghasoftenbeenregardedas dubi-ous,neitherUSPTOnorEPOpatentsmatterforvalueinanyofthesampleperiods.
administrativereformshave ledtoa strengtheningof thelegal
valueofpatents.IntheU.S.,akeystepinthisdirectionwasthe
cre-ationin1982ofaspecializedCourtofAppealsoftheFederalCircuit
(CAFC).AlthoughthestatedobjectiveofCongresswastoprovide
uniformityinpatentlitigationcases,thecourtsoondemonstrated
a“pro-patent”disposition(MazzoleniandNelson,1998;Merges,
1992).Specifically,thecourtwasfoundtoupholdinappealalarger
proportionofdecisionsfavorabletothepatentholder,toreverse
alargerproportionofdecisionsunfavorabletopatentholder,and
tosubstantiallyincreasetherateofpreliminaryinjunctions,
com-paredtotheprevioussystem(Gallini,2002;LanjouwandLerner,
2001).Thedefinitionofpatentablesubjectmatterwasalso
broad-enedtoincludeemerging newfieldssuchasbiotechnologyand
software,andpatentdurationwaslengthened.Severalcountries
followedtheU.S.example,whichledovertimetoanupward
har-monizationofintellectualpropertyprotectionstandards(Guellec
andvanPottelsberghedelaPotterie,2007;Lerner,2002).
The strengthening of the legal value of patents gave firms
powerfulincentivestoseekpatentprotection(HallandZiedonis,
2001;KortumandLerner,1999).Patentapplicationsnearlytripled
between1980and2000intheU.S.,withEuropeand Japan
fol-lowingsuit.Butaspatentsbecamemorepowerfullegalweapons,
concernsabouttheeasewithwhichtheyweregrantedbeganto
grow.Observersnotedthatwaytoooftenpatentsweregranted
forideasthatwereneithernovelnornon-obvious.Some–most
notably Jaffe and Lerner (2004)– went asfar asto argue that
lowerstandardscouldendangerinnovationandprogressbecause,
ifapplicantscouldgetpatentsforobviousorexistingideas,then
patentscouldusedtoharass,seekcompensationfrom,orevenshut
downlegitimatebusinesses.Overwhelmingly,however,criticism
aboutdecliningpatentabilitystandardshasbeendirectedtoward
theAmericansystem.Severalargumentsunderpinthebeliefthat
examinationstandardsarelower(and,untilquiterecentlyatleast,
havebeengettingworse)intheU.S.thaninEuropeorJapan.
First,theUSPTOisseverelyunderstaffedrelativetoothermajor
patentoffices.GuellecandvanPottelsberghedelaPotterie(2007)
estimatethat“boththeincomingworkloadofexaminers(number
ofclaimsfiledperexaminer)andtheiroutput(numberofclaims
grantedperexaminer)isthreetofourtimeshigherattheUSPTO
thanattheEPO”(p.201).ThusUSPTOexaminersspend
consid-erablylesstimethanEPOexaminersoneachclaim.Retentionof
experiencedpersonnelhasalsoprovedtobeparticularlydifficult
attheUSPTO,duetoacombinationofrelativelylowsalariesand
attractiveoutsideoptions.5GuellecandvanPottelsberghedela
Pot-terieconcludethatdecisionsattheEPOtendtobebasedonamore
carefulandexpertassessmentofthepriorartthanattheUSPTO,
whichmightexplainthemuchhigherrejectionratesinEurope.6
Second,USPTOexaminersappeartofacestrongerpressuresto
acceptratherthanrejectapplicationsthanexaminersinEuropeor
Japan.Onereasonisthat,bytransformingtheUSPTOina
user-fee funded entity, the 1990 Omnibus Act may have promoted
5AccordingtoMejerandvanPottelsberghedelaPotterie(2011),turnoverof employeesattheUSPTOisabout33%,comparedtojust3%attheEPO.
6QuillenandWebster(2001)estimatethatthepatentapprovalratefortheUSPTO whencorrectedforcontinuingapplicationsisintherangeof95–97%,substantially higherthanfortheEPOorJPO.JaffeandLerner(2004)alsoshowthatsuccessful USPTOapplicationshaverisenintheperiodbetween1987and1998twiceasmuch as‘important’inventionsthatweregrantedinallthreeoftheworld’smajorpatent jurisdictions(theUSPTO,EPO,andJPO).Theyarguethatthisfindingis“hardto explaininanymannerotherthandecliningstandardsintheU.S.PTO,producing anever-growingproportionofU.S.patentsthepatent-holdersthemselvesdidnot thinkmeritedpatentingelsewhere”(p.143).Finally,Gallini(2002)notesthatifthe timeseriesofKortumandLerner(1999)isextendedtoincludethesecondhalfofthe 1990s,KortumandLerner’sconclusionthattheU.S.didnotbecomeanincreasingly attractivedestinationforforeigninventorsistempered,thus“lendingsomesupport forthe‘friendlycourt’hypothesis”(p.138).
1498 S.Belenzon,A.Patacconi/ResearchPolicy42 (2013) 1496–1510
a“customer-friendly”mentality.JaffeandLerner(2004)provide
several examples illustrating the dysfunctional effect of such
a mentality. Most importantly,USPTO examiners are rewarded
accordingtohowfasttheyprocessapplications.However,since
applicantscanmodifyandappealinitiallyrejectedpatents,
rejec-tionstendtobeverytime-consuming.Thus,USPTOexaminershave
strongfinancialincentivestoacceptratherthanrejectapplications.
Third,fromitsinceptiontheEPOallowsthirdpartiesto
chal-lengeclaimsinapatentforninemonthsafterthepatentisgranted.
Thispermitsadditionalpriorartreferencestobesubmittedand
allowsfurtherverificationofthepatentabilityconditions(Mejer
andvanPottelsberghedelaPotterie,2011).IntheU.S.,by
con-trast,asimilarpost-grantreviewprocesshasonlyrecentlybeen
introducedwiththe2011AmericaInventsAct.
Fourth, until quite recently, court decisions have tended to
weaken U.S. patentability standards. The CAFC established, for
instance, that prior publication obviates novelty only when all
thefeaturesofaninventionhavebeenrevealedbyasingleprior
publication,7andestablished“thereasonableexpectationof
suc-cess”asalowerstandardofnon-obviousness8(Barton,2000;Lesser
andLybbert,2004).Therearealsofewerlimitationsonpatentable
subjectmattersintheU.S.thaninEurope.Inventionsintheareasof
software,businessmethodsandgenetics,forinstance,are
gener-allypatentableintheU.S.butnotinEurope.Theexplosionofpatent
activityinnewtechnologyareashasbeenviewedwithparticular
concernbycriticsoftheUSPTO.Examinationstandardsappearto
havebeentoolowinthesefields(atleastinitially),dueto
inexpe-rienceonthepartofexaminers,lackofadequatewrittenpriorart
documents,andcourtdecisions(Merrilletal.,2004;Hall,2007).
Lastbutnotleast,patentingcostsaremuchlowerintheU.S.than
inEuropeorJapan.vanPottelsberghedelaPotterieandFranc¸ois
(2009)estimatethat anEPOpatentthatis renewedfor20 (10)
yearsin13memberstatescostsalmost9(4.6)timesasmuchasa
USPTOpatent.Anobviouseffectoflowpatentingcostsisto
encour-agelow-valueapplicationsandincreasethevarianceofthepatent
qualitydistributionbystretchingitsbottomtail.Perversefeedback
effectscanalsoarise,withlowpatentingcostsexacerbating
work-loadproblems.9 Moreover,firmsmay,whentheyhaveachoice,
startfilingstrategically,byfirstapplyingforUSPTOpatentsand
onlysubsequently,iftheinventionsturnouttobevaluable,forthe
moreexpensiveEPOpatents.Combined,theseeffectscouldlead
overtimetoawiderqualitygapbetweenEPOandUSPTOpatents.10
2.2. PatentportfolioevaluationinM&A:theeffectofinstitutional change
Fromafinancialperspective,thefirmcanbeviewedasa
portfo-lioofinvestmentprojects;thusthevalueofthebusinessasawhole
canbecomputedasthenetpresentvalueofthecashflowsit
gener-ates.AsmentionedintheIntroduction,patents(andbyextension
patentportfolios)canbepositivelycorrelatedwithfirmvaluations
7StudiengesellschaftKohlev.DartIndus.Inc.,726F.2d724(Fed.Cir.1984).
8Amgen,Inc.v.ChugaiPharmaceuticalCo.,927F.2d1200(Fed.Cir.1991).
9Howchangesinthepatentfeestructurecanimprovethepatentquantity-quality tradeoffisanimportanttopicinpolicydebates,especiallyaftertheUSPTOraisedits feesin2004.SeeGuellecandvanPottelsberghedelaPotterieforadiscussion(2007, pp.212–213).
10Ingeneral,firmshavethepossibilitytoseekpatentprotectioninmultiple countries,butwillonlyusethispossibilityforinventionsthattheydeem suffi-cientlyvaluabletojustifytheextraexpense.USPTOandEPOpatentsareinthis respectarguablydifferentbecause,whileaUSPTOpatentmayormaynotbepartof apatentfamilycomprisinganumberofequivalentpatentsindifferentcountries,an EPOpatentisalmostalwaystheresultofadecisionthattheinventionisworththe expenseofseekingprotectioninmultiplecountries,asasinglenationalapplication wouldotherwisebelessexpensive(wethankananonymousrefereeforpointing thisouttous).
because(i)theyestablishexclusionaryrightsonpatented
inven-tionsand(ii)provideusefulsignalsaboutthequalityofpatented
inventionsandoverallcorporatetechnologicalstrength.
Severalstudiesconfirmtheusefulness ofpatent statisticsas
indicatorsoffirms’inventiveeffortsandvalue(e.g.,Griliches,1990).
Dengetal.(1999)findthatthenumberofpatentsgrantedtoa
firminagivenyearandpatentcitationmeasures(forward
cita-tionsand“sciencelink”)arestronglyassociatedwithsubsequent
market-to-book ratios.Positiveassociations betweenfirmvalue
and citation-weighed patent indexes are also found by Bloom
andVan Reenen(2002),Halletal. (2005)and Belenzon(2012),
among many others. Pakes (1985) demonstrates that an
unex-pected increase in the number of patents granted toa firm is
associatedwithalargechangeinmarketvalue.Usinga
comprehen-sivesurveyinstrument,Gambardellaetal.(2008)directlyaskfor
informationabouttheperceivedvalueofpatents,ratherthan
try-ingtoinferpatentvaluesusingestimationtechniques.Theyreport
veryhighestimates,withtheaveragepatentvalueaboveEUR10
million,andamedianofEUR650thousand.Lastly,inasampleof
pharmaceuticalfirms,Narinetal.(1987)findthatpatentcountsare
highlycorrelatedwithseveralmeasuresofcorporatetechnological
strengthsuchasthenumberofnewdrugsregistered,important
newdrugsapproved,scientificarticlespublished,andexpert
opin-ionsofcorporatetechnicalstrength.
Inthispaper,wefocusonthecontributionofpatentportfolios
tothevaluationoftargetfirmsinM&Adeals.Inevaluatingpatent
portfolios,acquiringfirmsmusttakeseveralfactorsintoaccount.
Factorsfrequentlymentionedbypractitionersinclude:(i)changing
marketconditionsandtechnologycycles,(ii)thelevelofnoveltyof
thepatentedinventions,(iii)thestrengthandbreadthof
exclu-sionaryrights,(iv)thedifficultyofinventingaround,(v)therisks
ofpatentdisclosure,(vi)afirm’soverallportfolioposition
(includ-ingthepotentialforfurtherpatents),and(vii)bargainingpotential
(Wilson,2007).Moreover,inM&A thevalueofpotential
syner-giesbetweenacquiringandtargetfirmscanbepartlyincorporated
intothetargetfirm’sprice(SluskyandCaves,1991).Thus,patent
portfoliovaluationsinM&Amayalsoreflectfirm-specificand
deal-specificfactors,notjustthevalueofthepatentportfolioinsome
absolutesense.
The previous discussion suggests that patent reforms have
affectedboththelevelofnoveltyofthepatentedinventionsand
thestrength and breadthofexclusionaryrights (points(ii)and
(iii)intheclassificationabove).Weputforwardtwopotentially
conflictinghypotheses.Thefirstonestatesthat,becausepatents
havebecomemorepotentlegalweapons,theeffectofbothUSPTO
andEPOpatentsoncorporatevaluationshasgrownstrongerin
recenttimes.
H1(Strengtheningofpatentprotectionrights). Therelationship
betweenfirmvalueandUSPTOandEPOpatentshasgrownstronger
inrecenttimes.
The second hypothesis states that, because patent quality
appearstohave declinedmore in theU.S.than in Europe,EPO
patentsmayhavebecomeovertimeamorereliableindicatorof
technologicalstrengththanUSPTOpatents.Thus,theeffectofEPO
patentsonacquisitionvalueinM&Ashouldhavebecomestronger
overtime,relativetotheeffectofUSPTOpatents.Notethatitis
pos-siblethatthequalityofUSPTOpatentsmayhavedeclinedsomuch
astocompletelyoutweighthepotentialbenefitsofstrongerpatent
protectionrights.Thus,incontrasttowhatsuggestedin
Hypothe-sis1,therelationshipbetweenUSPTOpatentsandfirmvaluemay
havegrownweaker,notstronger,overtime.
H2(DecliningUSPTOpatentquality). Overtime,EPOpatents
have become more strongly related to firm value than USPTO
Twoadditionalpointsareworthemphasizing.First,asdiscussed
above,patentqualityintheU.S.mayhavedeclinedmoreinnew
technologyareas suchas biotechnology, software and business
methodsthaninmoretraditionalareas(Barton,2000;Hall,2007).
Thus,changesintheimportanceofUSPTOpatentsovertimemay
notbeuniformacrosssectors.Wewilldistinguishbetween
differ-enttechnologysectorsintheempiricalpart.
Second,because theeconomic significance of patents varies
tremendously,simplepatentcountsmaybeverynoisyproxiesfor
thevalueofafirm’stechnology.Intheempiricalpart,wewilluse
forwardcitationstocontrolforthequalityofpatentedinventions.11
BecauseourdiscussionsuggeststhatUSPTOpatentsaremore
het-erogeneousintheirqualitythanEPOpatents(and,inparticular,
thattheyexhibitalongertailatthelowerendofthequality
distri-bution),weexpectcitation-weighingtostrengthenespeciallythe
effectofUSPTOpatents.
3. Data
Thispapercombinesdatafromthreesources:(i)M&Adatafrom
ThomsonSDCPlatinum,(ii)informationonpatentsfromtheUnited
StatesPatentandTrademarkOffice(USPTO),and(iii)information
onpatentsfromtheEuropeanPatentOffice(EPO).
WedropM&Adealsthatdonotreportinformationonvalue,net
totalassetsandacquiredstakes.Wealsodropdealswherethe
tar-getfirmisnotfromanOECDcountry,becausepatentlawmayvary
substantiallybetweendevelopedanddeveloping countries.This
leavesuswith30,306deals.40%ofthetargetsareAmerican,20%
areBritish,7%areJapanese,andtheremainingarefromotherWest
Europeancountries.Wematchthesefirmstoourpatentdatasets.
Priortotheacquisitioncompletionyear,4,643firmshaveatleast
onepatentintheUSPTOorEPO.
Table1summarizesdescriptivestatisticsforthemainvariables
usedinouranalysis.Firmvalueiscomputedastheratiobetween
transactionvalueandacquiredstakes.Theaveragetargetfirmis
valuedat$228million,has$118millionintotalassets,generates
$138millioninannualsales,andmakes$27millioninprofits.Two
separatepatentstocksareconstructed:oneforUSPTOpatentsand
anotherforEPOpatents.Weincludeonlypatentsthataregranted
priortotheacquisitioncompletionyear.Stocksarecomputedusing
theperpetual inventorymethodwitha depreciation rateof 15
percent.Tocontrolforpatentquality,wealsoconstruct
citation-weighedstocksbyweighingeachpatentbythenumberofcitations
itreceives(excludingself-citations)dividedbytheaverage
num-berofcitationsreceivedbyallotherpatentsthataregrantedbythe
samepatentofficeinthesameyear.Foreachpatent,weinclude
onlycitations fromotherpatentsthat aregranted bythesame
patentoffice.Thatis,forUSPTOpatents,weonlyincludecitations
fromUSPTOpatents,andforEPOpatentsweonlyincludecitations
fromEPOpatents.Finally,weconstructanaggregatepatent
mea-sure–totalpatentstock–whichissimplythesumofUSPTOand
EPOpatentstocksforeachfirm.Theaveragepatentingfirm(afirm
withatleastonepatentfromtheUSPTOorEPO)hasastockof14
USPTOpatents(amedianof3),and8EPOpatents(amedianof1).
11 SincethelandmarkworkofTrajtenberg(1990),citation-weighedpatentindexes havebeenshowntobestronglycorrelatedwiththesocialvalueofinnovations (Trajtenberg,1990),peerevaluationoftheirtechnicalimportance(Albertetal., 1991),renewal decisions(Harhoffetal.,1999;Thomas,1999),andfirmvalue (Belenzon,2012;BloomandVanReenen,2002;Dengetal.,1999;Halletal.,2005; HirscheyandRichardson,2004).Halletal.(2005),inparticular,findthatinvestors areabletoaccuratelyforecasttheexpectedvalueofpatentedinventions,asitis laterconfirmedbyfuturecitations.Analternativewaytocontrolforpatentquality wouldhavebeentoestimatetheprivatevalueofpatentsusingannualrenewalfees, asinSchankermanandPakes(1986).
Since ourfocus is onhow therelationship between
patent-ingandacquisitionvaluehaschangedovertime,Table2presents
descriptivestatisticsacrosstimeperiods.Thesampleincludesboth
patentingandnon-patentingtargetfirms.Wesplitthesampleinto
sixtimeperiods.Averagefirmvalueisquitestableovertime.
Aver-agevalueis$233millionfor thedealscompletedintheperiod
1985–1990,and$249millionforthedealscompletedintheperiod
2005–2007.Ontheotherhand,thereisasubstantialriseinTobin’s
q–theratiobetweenvalueandassets–overtime.AverageTobin’s
qis3.4fordealscompletedin1985–1990,andis4.9fordeals
com-pletedin 2005–2007.Thus intangibleassetssuchasknowledge
appeartohavebecomemoreimportantovertime.
Table3examineshowkeycharacteristicsoftargetfirmshave
changedovertime.Onaverage,92%ofpatentingfirmshaveatleast
oneUSPTOpatent,and72%haveatleastoneEPOpatent.Overtime,
theshareoffirmsthatpatentintheUSPTOdeclinesbutremains
veryhighwhiletheshareoffirmsthatpatentintheEPOincreases
substantiallyfrom56%in1985toabout77%in2007.Thereisa
sharpincreaseinacquisitionvalueintheperiod1995–1998,but
thesehighvaluessubsequentlyreturntotheoverallsamplemean.
Thus,thereisnoclearpatternregardinghowfirmvalueinthese
acquisitiondealshaschangedovertime. Tobin’sq,ontheother
hand,increasessubstantiallyovertime,albeitinanon-monotonic
way(italsopeaksin1998).Theshareofnon-Americanacquisitions
alsochangessubstantiallyovertime(notreportedinatable).Inthe
wholesample40%oftheacquisitionsinvolveAmericantargets,yet
thisfiguresfallsfromaround69%in1985toaboutaquarterofthe
dealsin2007.Wediscusshowthisissuemayaffectthe
interpre-tationofourfindinginthenextsectionandwhenpresentingthe
estimationresults.
Afactthatemergesclearlyfromourstatisticsisthatvaluations
wereveryhighduringthe1995–1998period.Themostlikely
expla-nationforthesehighvaluationsistheexuberancethataffected
investorsandstockmarketsduringthedot-combubbleyears.This
bubblestartedinMarch1995withtheexcitementoccasionedby
Netscape’sinitialpublicoffering,and terminated in 2000–2001
withthecollapseoftheNASDAQCompositeindex(however,the
Americaneconomystartedlosingspeedsince1999).Toaccountfor
possiblyinflatedcorporatevaluationsduringthisperiod,intherest
oftheanalysis(asinTable1),wewillreportseparateestimations
fortheperiods1995–1998and1999–2001,whichcorrespondto
theboomandbustofthedot-combubble.
4. Econometricanalysis
Weestimatethefollowingspecificationforthevalueoftarget
firms,bothforthewholesampleandfordifferenttimeperiods.12
lnValueit =ˇ0lnAssetsit+ˇ1ln(1+PatUSit−1)
+ˇ2ln(1+PatEUit−1)+t+i+cc+εi. (1)
Valueitis thevalueof targetfirmiatacquisitioncompletion
yeartandiscomputedastransactionvalueoveracquiredequity.
Assetsitistotalnetassets.PatUSit−1andPatEUit−1are,respectively,
targetfirmi’sUSPTOandEPOpatentstockspriortothe
acquisi-tioncompletionyear.t,iandccarecompletesetsofdummiesfor
acquisitioncompletionyears,twodigittargetindustrySICcodes,
andtargetcountry.εiisaniiderrorterm.Ourmaininterestisthe
waythecoefficientsˇ1andˇ2changeovertime.Weinvestigate
thisintwoways:byinteractingUSPTOandEPOpatentstockswith
perioddummies,andbyestimatingseparatespecificationsfor
dif-ferenttimeperiods.BecausechangesinthequalityofUSPTOand
1500 S.Belenzon,A.Patacconi/ResearchPolicy42 (2013) 1496–1510
Table1
Summarystatisticsformainvariables.
Variable #acquisitions Mean Std.dev. Distribution
10th 50th 90th Targetvalue($,mm) 30,306 228 1,102 5 60 576 Netassets($,mm) 30,306 118 1,890 7 30 281 Tobin’sq($,mm) 30,306 4.0 4.9 0.7 2.1 10.2 Sales($,mm) 26,245 138 197 4 52 416 Profits($,mm) 21,714 27 64 −2 7 79 %ofshareacquired 30,306 0.58 0.42 0.04 0.71 1
Totalpatentstock,count 4,643 22 194 0.7 4 42
Totalpatentstock,cite 4,643 21 161 0.4 4 43
USPTOpatentstock,count 4,643 14.4 142.9 0 2.6 25.5
USPTOpatentstock,citation-weighed 4,643 13.9 98.6 0.2 2.0 27.9
EPOpatentstock,count 4,643 7.7 54.3 0.5 1.1 14.5
EPOpatentstock,citation-weighed 4,643 7.5 64.8 0.3 0.9 13.5
Notes:Thistablereportssummarystatisticsforthemainvariableusedintheestimation.ThesampleincludesM&AtransactionsfromSDCPlatinumbetweenintheperiod 1985–2007.Firmvalueistheratiobetweentotaldealvalueandacquiredequity.Tobin’sqistheratiobetweenfirmvalueandassets.Patentvaluesarereportedforfirms withatleastoneUSPTOorEPOpatent.Weweighpatentsbycitationsbydividingthenumberofcitationstheyreceivebytheaveragenumberofcitationsreceivedbyall patentsthataregrantedbythesamepatentofficeinthesameyear.Robuststandarderrorsareinbrackets.
Table2
Summarystatisticsformainvariablesfordifferenttimeperiods.
Variable 1985–1990 (4,770deals) 1991–1994 (6,287deals) 1995–1998 (6,287deals) 1999–2001 (4,890deals) 2002–2004 (2,694deals) 2005–2007 (5,420deals) Mean Median Mean Median Mean Median Mean Median Mean Median Mean Median
Firmvalue($,mm) 233 72 194 37 222 64 262 73 204 56 249 74 Netassets($,mm) 108 37 154 19 92 31 103 33 110 38 126 29 Tobin’sq($,mm) 3.4 2.1 3.7 1.9 3.9 2.1 4.4 2.1 3.0 1.5 4.9 2.7 Sales($,mm) 167 74 123 40 120 42 143 55 146.4 60.3 141.8 58.2 Profits($,mm) 42 15 23 5 27 7 27 8 22.9 5.2 23.5 5.4 %ofshareacquired 42.9 78.1 61.3 56.3 54.7 55.3 1.0 65.7 56.5 51.3 58.2 62.3
USPTOpatentstock,count 18.5 3.9 12.4 3.1 10.5 2.7 11.9 2.0 29.4 1.9 9.8 1.6
USPTOpatentstock,citation-weighed 18.3 3.8 14.0 2.5 13.0 2.2 11.0 1.4 21.6 1.1 8.0 1.1
EPOpatentstock,count 7.3 0.4 8.1 1.4 6.0 1.2 7.5 1.0 12.0 1.2 7.5 1.3
EPOpatentstock,citation-weighed 6.9 0.3 7.6 1.1 5.5 1.0 7.1 0.8 13.5 1.0 7.4 1.2
Notes:Thistablereportssummarystatisticsforthemainvariableusedintheestimationbymainperiods.ThesampleincludesM&AtransactionsfromSDCPlatinumbetween intheperiod1985–2007.Firmvalueistheratiobetweentotaldealvalueandacquiredequity.PatentdatafromtheUSPTOcovertheperiod1975–2007,andpatentsdata fromtheEPOcovertheperiod1979–2007.Weweightpatentsbycitationsbydividingthenumberofcitationstheyreceivebytheaveragenumberofcitationsreceivedby allpatentsgrantedinthesameyear.
Table3
Patentingfirmsacrossyears.
Year #deals Firmvalue Tobin’sq %PatentintheUSPTO %PatentintheEPO
1985 52 388.1 2.3 100.0 55.8 1986 98 510.8 3.0 98.0 62.2 1987 134 322.6 3.1 98.5 62.7 1988 187 319.1 2.4 95.2 62.0 1989 227 381.2 3.1 95.2 66.1 1990 216 393.4 3.1 94.0 72.2 1991 189 275.7 2.9 91.5 69.8 1992 186 278.8 3.0 94.1 75.8 1993 222 278.0 3.9 89.2 73.9 1994 280 254.9 3.6 94.3 71.4 1995 253 466.7 4.2 96.0 71.1 1996 282 312.6 3.9 94.3 69.5 1997 285 390.3 5.0 96.1 73.3 1998 86 639.5 4.3 94.2 67.4 1999 336 275.3 4.2 90.8 73.8 2000 211 306.2 4.1 92.4 73.5 2001 137 213.0 3.3 93.4 74.5 2002 229 296.4 2.9 90.8 73.8 2003 203 179.8 2.8 91.6 77.8 2004 30 598.9 4.0 93.3 73.3 2005 246 308.0 4.6 85.0 76.8 2006 357 342.7 3.9 82.6 78.2 2007 197 358.1 3.6 83.2 76.6 Total 4,643 329.8 3.6 91.9 72.1
Notes:Thistablereportsthenumberofpatentingfirmsforeachyearinoursample.WeincludeonlyfirmsthathaveatleastonepatentinUSPTOorEPObytheiracquisition year.Firmvalueistheratiobetweentotaldealvalueandacquiredequity.Tobin’sqistheratiobetweenfirmvalueandassets.%patentintheUSPTO(EPO)isthepercentage offirmsthathaveatleastonepatentintheUSPTO(EPO)outofallpatentingfirmsinthesameyear.
EPOpatentsmayinpartdriveourresults,wewillalsocomparethe estimatesˇ1andˇ2forspecificationswherepatentsareweighed bycitationsandspecificationswherepatentsarenotweighedby citations.13
Thereareanumberofimportantpotentialbiasesthatmayaffect
ourestimationswhichareworthdiscussingbeforepresentingthe
results.
Afeatureofoursampleisthatitsfirmcompositionchangesfrom
oneyeartoanother.Thisleavesourresultssensitivetounobserved
heterogeneitybias.Aconcernmightbe,forinstance,thatfirmsthat
choosetopatentintheEPOinthelatesampleperiodmaybe,on
average,ofhigher(unobserved)qualitythanfirmsthatpatentin
theEPOinthebeginningofthesampleperiod.Thismeansthat
laterinthesampleperiodthecorrelationbetweenEPOpatentsand
unobservedfirmqualityispositive,causinganupwardbiasinˇ2.
Wecancheckwhetherthisconcernislikelytoaffectourestimates
bycomparingobserveddealcharacteristicsforearlyandlaterEPO
patentees.ForfirmsthatpatentintheEPOinthefirstestimation
period,1985–1990,averagedealvalueis$395million,Tobin’sqis
2.9,salesare$242million,andprofitsare$63million.Forfirmsthat
patentintheEPOinthelastsampleperiod(2005–2007),average
valueislowerat$370million,butTobin’sqishigherat4.2.Salesare
similartotheearlyperiodlevelat229,butprofitsaremuchlower
at$33million.Fromthiscomparison,itisnotclearwhetherearlier
EPOpatenteesareoflowerqualitythanlaterones.Interestingly,we
dofindthatthatlaterEPOpatenteesbenefitfromhigherTobin’sqas
comparedtoearlyEPOpatentees.Thisisconsistentwiththeview
thatknowledgeassetsforthesefirmsaccountforagreaterportion
offirmvalue.Ourestimationresultswhicharepresentedbeloware
consistentwiththisinterpretation.
Anotherpotentialsourceofsampleselectionisindustry
com-position. In the later period of the sample we observe more
biotechnologyandinformationtechnologyacquisitionsthanin
ear-lierperiods.Iftheimpactofpatentsonfirmvaluationvariesacross
industries, then ourkey estimates could bebiased. Controlling
forindustryisalsoimportantbecause,asourdiscussionsuggests,
patentqualitymayhavedeclinedmoreinsomeindustriesthan
others.Tocheckthesensitivityofourresultstoindustry
varia-tion,wereportourmainestimationresultsseparatelyforthemain
innovativeindustriesinoursample.14
Lastly,weobserveasharpdeclineintheshareofAmerican
tar-getsinoursample.Whileintheearlyperiodsabout50percentof
targetsareAmerican,thispercentagedropstoabout20percentin
thelatesampleperiods.Asimilarpatternemergeswhenlookingat
patentingfirmsonly(from60to30percent).Thiscouldaffectour
results,forinstanceifUSPTOpatentsmattermostlytoAmerican
firms,butalsonon-AmericanfirmspatentattheUSPTO.Thus,when
presentingtheeconometricresults, wealsodiscuss the
robust-nessofourfindingstosplittingthesamplebetweenAmericanand
non-Americantargets.
13 Animportantcaveatisinorder.Apremiseofouranalysisisthatpatentcitations areausefulproxyforpatentquality.However,becausethecitationsapatentreceives accumulateovertime,patentqualityislikelytobemeasuredwitherrortowardthe endoftheestimationperiod.IftheEPOprovidessomequalityreassurancebecause few“bad”patentsaregrantedbytheEPO,thenˇ2islikelytobehigher(andˇ1lower) inthelastyearsoftheestimationperiodevenifpatentsareweightedbycitations, simplybecauseofcitation-truncation.Yet,findingthatUSPTOpatentsbecomeless valuablebeforetheendoftheestimationperiod(aswedo,seeTable5,columns4–5 and10–11)suggeststhattruncationisnotlikelytocompletelydriveourresults.
14 Wealso checked how the representation of different industries changes throughoutoursampleperiod.Wefocusedondrugsandchemicalsbecause patent-ingisveryimportantintheseindustries;thuschangesinrepresentationovertime havethepotentialtoaffectourestimates.Wedidnotfindsubstantialvariationin therepresentationoftheseindustriesinoursampleovertime.In1985,4.6%ofthe dealsareindrugsandchemicals.In2007,theproportionis5.0%.
5. Estimationresults 5.1. “Horse-race”estimation
Table4reportstheestimationresultsforthecompletesample
period.Westartwithpatentmeasuresnotweighedbycitations
(thus,weusethenumberofpatentseachfirmholds when
con-structingtheEPOandUSPTOpatentstocks).Column1includes
loggedtotalpatentstock(thelogofthesumofUSPTOandEPO
patentstocks). Theelasticityestimateof acquisitionvalue with
respecttototalpatentstockis0.117(astandarderrorof0.007).
Obtaininganelasticityestimateallowsustocalculatetheincrease
inacquisitionvalueasaresponsetoaunitaryincreaseinpatent
stock.Forthiscomputation,weusetheaverageacquisitionvalue
andaveragepatentstockvaluesforpatentingfirmsinthesample.
Theelasticityestimatefromcolumn1impliesthatincreasingtotal
patentstockbyoneisassociatedwitha$1.2millionincreasein
acquisitionvalue.
Columns2and3addthestocksofUSPTOandEPOpatents
sep-arately.Inbothcasesthecoefficientestimatesarelargeandhighly
significant.Incolumn4weestimatea“horse-race”betweenUSPTO
andEPOpatents.Includingbothmeasuresinasingleregression
resultsinadropinbothcoefficientestimatesascomparedtowhen
eachvariableisseparatelyincluded.Whenbothstocksareincluded,
theelasticityofacquisitionvaluewithrespecttoEPOpatentsis
almosttwiceaslargeastheelasticitywithrespecttoUSPTOpatents.
Basedontheestimatesfromcolumn4,raisingtheEPOpatentstock
byoneunitisassociatedwitha$2.6millionincreaseinfirmvalue,
comparedtoa$0.8millionincreasefortheUSPTOpatentstock.
Columns 5–7 report initialrobustness checks for the above
results.Incolumn5weexcludedealsininformationtechnology
andbusinessservicesbecausepatentabilityrequirementsinthese
fieldsvarysubstantially betweentheU.S.and Europe.The
esti-matesarenotsensitivetoremovingtheseindustries.Incolumn
6weincludeonlypatentingfirmstoensurethatourresultsarenot
drivenbycomparingpatentingtonon-patentingfirms.Ifselection
intopatentingiscorrelatedwithanomittedvariablethatisrelated
tovalue,thenourpatentingvariableswouldbeupward-biased.We
findthesamepatternofresultswhenfocusingonlyonfirmswith
atleastonepatent.Incolumn7weincludeonlydealsthatare
clas-sifiedas“high-tech”inSDC.Ourconcernhereisthatourestimates
maybedrivenbycomparinghigh-techdeals(wherepatent
valua-tionsmaybehigh)tolow-techdeals(wherepatentvaluationsmay
below).Theestimatesremainstable.
Columns8–14presenttheestimationresultswhencontrolling
forpatentqualityusingforwardcitations.Controllingforcitations
proves importantfor USPTO patents,but much less sofor EPO
patents.ThisisexpectedundertheassumptionthatUSPTOpatents
areofgreaterqualityheterogeneitythanEPOpatents.Controlling
forcitationsincreasesthecoefficientestimateforUSPTOpatents
bynearly30%(from0.056to0.072).Ourestimatesimplythata
unitaryincreaseincitation-weighedEPOandUSPTOpatentstocks
is associated with$2.3and $1.1 millionincrease in acquisition
value,respectively.These estimatesstill implylargedifferences
inthevalueofUSPTOandEPOpatents;however,thedifference
isconsiderablylowerthanwhenwedonotcontrolforcitations.
AnotherinterestingresultistherobustnessoftheUSPTO
coeffi-cientincolumns12–14wherewe performthesamesensitivity
checksasreportedincolumns5–7.
Comparingourfindingstopreviousestimatesischallengingdue
totheheterogeneityof methodsused andthefact that weare
notawareofotherstudiesthatfocusonacquisitionvalue.These
caveatsnotwithstanding,weprovideafewcomparisonsto
lead-ingstudiesinthefield.First,we compareourestimatestoHall
et al.(2005).Like us, theyusefirm valuetoinfer patentvalue
1502 S. Belenzon, A. Patacconi / Research Policy 42 (2013) 1496 – 1510 Table4
Patentsandfirmvalue.
Dependentvariable:ln(Firmvalue)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Patentcount(notweighedbycitations) Patentsweighedbycitations
All All All All Exc.ITfirms Atleastonepatent OnlyHigh-Tech All All All All Exc.ITfirms Atleastonepatent OnlyHigh-Tech ln(1+TotalPatentStock)t−1 0.117** 0.122**
(0.007) (0.007) ln(1+USPTOPatent Stock)t−1 0.117** 0.056** 0.056** 0.048** 0.050** 0.125** 0.075** 0.076** 0.072** 0.072** (0.008) (0.012) (0.012) (0.015) (0.012) (0.008) (0.011) (0.012) (0.012) (0.012) ln(1+EPOPatent Stock)t−1 0.148** 0.100** 0.093** 0.090** 0.097** 0.152** 0.086** 0.079** 0.080** 0.080** (0.010) (0.014) (0.014) (0.014) (0.014) (0.010) (0.014) (0.014) (0.014) (0.014) ln(Netassets) 0.713** 0.714** 0.715** 0.713** 0.727** 0.716** 0.710** 0.712** 0.714** 0.715** 0.713** 0.727** 0.709** 0.709** (0.004) (0.004) (0.004) (0.004) (0.005) (0.012) (0.005) (0.004) (0.004) (0.004) (0.004) (0.005) (0.005) (0.005)
Two-digitindustrydummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Countrytargetdummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Acquisitionyeardummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
R2 0.680 0.679 0.679 0.679 0.689 0.683 0.653 0.680 0.679 0.679 0.680 0.690 0.653 0.653
Observations 30,306 30,306 30,306 30,306 26,532 4,643 24,010 30,306 30,306 30,306 30,306 26,532 24,010 24,010
Notes:ThistablereportstheresultsofOLSregressionsthatexaminetherelationshipbetweenpatentstocksandfirmvalue.ThesampleincludesM&AtransactionsfromSDCPlatinumbetweenintheperiod1985–2007.Patent
datafromtheUSPTOcoverstheperiod1975–2007,andpatentsdatafromtheEPOcoverstheperiod1979–2007.InColumns8–14weweighpatentsbycitations.Wedividethenumberofcitationstheyreceivebytheaverage
numberofcitationsreceivedbyallpatentsgrantedbythesamepatentofficeinthesameyear.Columns5and12excludefirmstheoperateinfieldsofinformationtechnologiesorbusinessservices.Robuststandarderrorsarein
brackets.
*Significantat5%.
data).Theyreportacitationselasticityofabout0.03,whilewefind
amuchhigher(citation-weighed)elasticityof0.122.Itis
impor-tanttonote,however,thatHalletal.(2005)alsocontrolforR&D
–informationwhichwedonothave.Thus,ourestimated
elastic-ityislikelytobepickingtheR&Deffectaswell(forinstance,an
increaseinnon-patentedknowledge).Interestingly,ourestimates
becomemuchclosertothoseofHalletal.(2005)when
consider-ingthehighly-citedpatentsintheirsample,andactuallybecome
lowerthantheirestimatesforfirmswithextremelylarge
num-bersofcitationsperpatent.Ourestimatesarealsomuchhigher
thatpreviousestimatesobtainedfrompatentrenewaldata.For
example,usingrenewaldataonEuropeanpatents,Pakes(1986)
andSchankerman andPakes (1986)reportvery lowvalue
esti-matesrangingfrom$847to$19,124(in1980prices).However,the
renewalapproachonlycapturesthe“patentpremium”–the
incre-menttothevalueofaninnovationwhichisrealizedbypatenting
it–nottheoverallvalueofapatentedinvention.Thelatteriswhat
wemeasure.Estimatesusingrenewaldatamayalsobedownward
biasedbecauseextremelyvaluablepatentsmaynotbeproperly
accountedfor(Harhoffetal.,2003;Aroraetal.,2008).Ourestimates
aremuchclosertothoseobtainedbyGambardellaetal.(2008)
usingthecomprehensivePATVALsurveyonthevalueofEuropean
patents.ThePATVALsurveydirectlyaskedforinformationofthe
perceivedvalueofpatents,ratherthantryingtoinfervalueusing
differentestimationtechniques.Theauthorsreportveryhighvalue
estimates,withaveragepatentvalueofaboveEUR10million,and
amedianofEU650thousand.Ourestimatesarequiteclosetotheir
surveymedianvalue.
5.2. Variationovertime
Next,westudyhowthevalueofUSPTOandEPOpatentshas
changedovertime.Forthispurposeweestimateseveral
“horse-race”specificationsfordifferentperiodsinoursample.Thetime
periodsareconstructedtryingtoeventhedistributionofnumberof
dealsovertime,whilealsoisolatingsomespecifichistoricalfactors
(thedot-combubblein1995–1998anditsburstin2000–2001)that
areevidentinourdescriptivestatistics.
Table 5 presents the estimation results. Two specifications
areprovided: flexible specificationswhere we splitthe sample
byperiods, and specifications where we include an exhaustive
setofinteractionsbetweentheperioddummiesandUSPTOand
EPO patent stocks. We begin with the case when patents are
notweighedbycitations(columns1–6).Aclearpatternemerges.
USPTOpatentsareimportantinthefirstsampleperiod1985–1990,
butsubsequentlyhavenosignificanteffectonacquisitionvalue,
exceptthaninthedot-combubbleyears1995–1998.Bycontrast,
theeffectofEPOpatentsisalwayslargeandsignificantfrom1991
to2007(columns1–5).
Estimationusingapooledregression(column6)confirmsthese
results.EPOpatentshaveastableeffectonvalue:wecannotreject
thehypothesisthattheEPOperiodinteractionsaredifferentfrom
oneanother(p-value=0.82).Yet,wecanstronglyrejectthe
hypoth-esisthattheyarejointly equaltozero(p-value<0.001). Onthe
otherhand,USPTOpatentsdonothavearobusteffecton
acqui-sitionvalue, andstarting from1991, withtheexception ofthe
1995–1998period,theireffectonacquisitionvalueisstatistically
zero.TheimportanceofEPOpatents,relative toUSPTOpatents,
clearlyincreasesovertime.Inthefirstsampleperiod,the
differ-encebetweenEPOandUSPTOcoefficientestimatesis0.017.The
samedifferenceinthefinalsampleperiodis0.120.
Asimilarpatternemergeswhenpatentsareweighedby
cita-tions(columns7–12).USPTOpatentsmatterforvalueinthefirst
three sample periods (1985–1998), but their effect completely
disappears in the 1999–2007 period. In contrast, EPO patents,
whichdonothaveasignificanteffectonvalueinthefirstperiod
0.16 0.18 0.2 0.08 0.1 0.12 0.14 0.16 EPO average 0.39 0.45 0.43 0.43 0.32 0.46 0 0.02 0.04 0.06 USPTO average 0.2 1985-1990 1991-1994 1995-1998 1999-2001 2002-2004 2005-2007 Share number of EPO patents USPTO patents EPO patents
a
b
0.12 0.14 0.16 0.18 0.2 0 37 0.60 0.44 0.41 0.50 0.04 0.06 0.08 0.1 USPTO average EPO average 0.37 0.35 0.41 0 0.02 1985-1990 1991-1994 1995-1998 1999-2001 2002-2004 2005-2007 Share number of EPO patents USPTO patents EPO patents Fig.1.(a)Time-varyingcoefficientestimatesonUSPTOandEPOpatentstocks. Patentsarenoweighedbycitations.(b)Time-varyingcoefficientestimatesonUSPTO andEPOpatentstocks.Patentsarenotweighedbycitations.(1985–1990),becomethedominanttechnologyindicatorinthelate
sampleperiods(1999–2007).15
Comparingweighedandun-weighedspecificationsshowsthat
citation-weighingbolsterstheeffectoftheUSPTOpatentsinthe
earlysampleperiods.Thatiswhatonewouldexpectifthe
qual-ityofUSPTOpatentsisveryheterogeneous.However,eveninthe
citation-weighedspecifications,thecoefficientonUSPTOpatents
dropstozerointhelatesampleperiods.Apossibleexplanationis
thatbecausecitationsaccumulateovertime,patentqualitymaybe
measuredwithlargeerrorneartheendofthesampleperiod.Thus
citation-weighingmaynotbeaparticularlygoodwaytocontrol
forpatentqualityinthelastfewyearsofoursample.Theeffectof
EPOpatents,bycontrast,alwaysremainsverystrong,bothinthe
un-weighedandweighedregressions.Fig.1graphicallyillustrates
theresultsfromcolumns6and12.
Overall,theresultsdonotsupportthehypothesisthatstronger
patentprotectionrightsledtohighervaluationsforUSPTOandEPO
patents(Hypothesis1).ThereissomeevidencethatEPOpatents
becamemorevaluableinthelasttwosampleperiods(2002–2004
and 2005–2007); however, USPTO patent coefficients certainly
didnotrise.Theresultsappear tobemore consistentwiththe
15Totalpatentsstock–thesumofUSPTOandEPOstocks–hasaverystableeffect overtime(notpresentedinthetable).Thecoefficientestimateonthe (citation-weighed)totalpatentsstockintheperiod1985–1990is0.102,comparedtoavalue of0.122inthe2002–2007period.
1504 S. Belenzon, A. Patacconi / Research Policy 42 (2013) 1496 – 1510 Table5
Patentsandfirmvalueovertime.
Dependentvariable:ln(Firmvalue)
(1) (2) (3) (4) (5) (5) (6) (7) (8) (9) (10) (11) (12) (13)
Patentcount(notweighedbycitations) Patentsweighedbycitations
1985–1990 1991–1994 1995–1998 1999–2001 2002–2004 2005–2007 All 1985–1990 1991–1994 1995–1998 1999–2001 2002–2004 2005–2007 All ln(1+USPTOPatent Stock)t−1 0.073** 0.033 0.103** 0.032 0.032 0.003 0.084** 0.064** 0.125** 0.056 0.041 0.014 (0.018) (0.030) (0.030) (0.036) (0.052) (0.034) (0.017) (0.028) (0.030) (0.033) (0.047) (0.034) ln(1+EPOPatent Stock)t−1 0.056 0.125** 0.087** 0.107** 0.134* 0.149** 0.043 0.103** 0.069* 0.094** 0.138* 0.143** (0.023) (0.033) (0.030) (0.041) (0.063) (0.039) (0.024) (0.033) (0.031) (0.038) (0.059) (0.039)
ln(1+USPTOPatentStock)t−1×
1985≤Year≤1990 0.046** 0.055** (0.017) (0.016) 1991≤Year≤1994 0.036 0.059* (0.030) (0.027) 1995≤Year≤1998 0.169** 0.185** (0.028) (0.028) 1999≤Year≤2001 0.053 0.099** (0.033) (0.031) 2002≤Year≤2004 −0.003 −0.000 (0.048) (0.044) 2005≤Year≤2007 −0.012 −0.000 (0.033) (0.033)
ln(1+EPOPatentsStock)t−1×
1985≤Year≤1990 0.063** 0.053* (0.024) (0.025) 1991≤Year≤1994 0.131** 0.114** (0.033) (0.032) 1995≤Year≤1998 0.086** 0.072** (0.030) (0.031) 1999≤Year≤2001 0.092* 0.061** (0.040) (0.038) 2002≤Year≤2004 0.139** 0.144** (0.059) (0.054) 2005≤Year≤2007 0.118** 0.110** (0.038) (0.037) ln(Netassets) 0.734** 0.711** 0.726** 0.711** 0.751** 0.674** 0.713** 0.734** 0.711** 0.725** 0.711** 0.751** 0.674** 0.713** (0.011) (0.009) (0.010) (0.010) (0.014) (0.010) (0.004) (0.011) (0.009) (0.010) (0.010) (0.014) (0.010) (0.004)
Two-digitindustrydummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Countrytargetdummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Acquisitionyeardummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
R2 0.726 0.693 0.686 0.671 0.683 0.664 0.680 0.726 0.693 0.686 0.671 0.683 0.664 0.680
Observations 4,770 6,287 6,245 4,890 2,694 5,420 30,306 4,770 6,287 6,245 4,890 2,694 5,420 30,306
Notes:ThistablereportstheresultsofOLSregressionsthatexaminetherelationshipbetweenpatentsandfirmvalueindifferenttimeperiods.Robuststandarderrorsareinbrackets.
*Significantat5%.
hypothesisthatUSPTOpatentqualityandvaluationsdeclinedover
time(Hypothesis2).
Usingtheestimatesfromcolumn13, wecanagaincalculate
patentvaluesfordifferenttimeperiodsseparatelyforUSPTOand
EPOpatents.Inthefirstestimationperiod,1985–1990,aunitary
increaseinUSPTOpatentstockleadstoa$765thousandincrease
inacquisitionvalue.Inthethirdperiod,1995–1998,whereUSPTO
patentsseemtobethemostvaluable,thesameunitaryincrease
leadstoa$2.8millionriseinacquisitionvalue.Thisvalue,
how-ever,dropstozerointhe2002–2007period.EPOpatentsexhibit
amuchmorestablecorrelationwithacquisitionvalue.Inthefirst
estimationperiod,1985–1990,aunitaryincreaseinthestockofEPO
patentsleadstoa$1.4millionincreaseinvalue.Thisvaluerisesto
$3.1millioninthesecondperiod1991–1994,whichisthesame
valuethatisobtainedinthelastestimationperiod(2005–2007).
Inunreportedspecifications,werepeatedtheanalysisinTable5
forpatentingfirmsonly,forfirmsclassifiedashigh-techbySDC,
andexcludingITandbusinessservices.Thesamepatternofresults
continuestohold.16
6. Robustness
Thepreviousanalysis uncoveredtwo important findings:(i)
overtime,EPOpatentsbecomemorevaluablerelativetoUSPTO
patents,and(ii)theeffectofUSPTOpatentsonacquisitionvalue
completelydisappearsinthelatesampleperiods.Inthissection
weexploretherobustnessofthesefindingsbysplittingthesample
accordingtodifferentcriteria.Duetothefewerdegrees-of-freedom
associated with suchexperimentation, the sample is split into
fewertimeperiods.
6.1. Targetfirmsize
Patentsplayaparticularlyimportantroleinthestart-upphaseof
afirm’slife(Long,2002).Theyprotectyoung,smallfirmsfrom
com-petitionbyindustryleaders(Teece,1986).Topotentialinvestors,
they provide “evidence that the company is well managed, is
ata certain stage in development,and hasdefined and carved
outamarketniche”(Lemley,2001,pp.1505–1506).Not
surpris-ingly,therefore,venturecapitalists,businessangelsandcorporate
investorsincreasinglybasetheirinvestmentdecisionsonIP
con-siderations(KortumandLerner,2000).
Table6examinestheextenttowhichourresultsaredrivenby
differencesintargetfirmsize.Weclassifyafirmassmallifitssales
fallinthelowestquartileofthesalesdistribution,andaslargeif
itssalesfallinthehighestquartile.Consistentwiththeideathat
patentsmatterthemostforsmallfirms,ourresultsholdonlyin
thesmall-firmsample.For largefirmswefindnoevidencethat
patents(eitherUSPTOorEPO)matterforvaluation,andthereisno
importanttemporalvariationintheirimportance.Bycontrast,EPO
patentsstronglycorrelatewiththevaluationofsmallfirms,
espe-ciallyinthelastsampleperiodwhentheyarealmostasimportant
forvaluationasnetassets.TheeffectofUSPTOpatentsonsmall
16 WealsomorecloselyexaminetheriseinUSPTOpatentvaluesduringthebubble years1995–1998.TheriseinUSPTOpatentvaluesisrobusttoremovinganyyear inthisperiod.Forinstance,whenremovingyear1998(anoutlierintermsofdeal valuesasshowninTable3),thecoefficientestimateonUSPTOpatents (citation-weighed)is0.125andthecoefficientestimateonEPOpatentsis0.063.Splittingthe samplebyAmericanandnon-AmericantargetsshowsthattheriseinUSPTOpatent valuesisdrivenbynon-U.S.targets.ForAmericantargets,theUSPTOcoefficient estimateis0.070;thesameestimatefornon-Americantargetsis0.227.Lastly,the coefficientestimateonUSPTOpatentsisabout50%largerinthe1995–1998period fortargetsintheinformationtechnologyfieldthaninalltheotherindustriesinthe
sameperiod. Table
6 Low sales vs. high sales firms. Dependent variable: ln( Firm value ) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) Patent count (not weighed by citations) Patents weighed by citations Sales ≤ 25th percentile Sales > 75th percentile Sales ≤ 25th percentile Sales > 75th percentile All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 ln(1 + USPTO Patent Stock )t− 1 0.072 ** 0.136 ** 0.059 − 0.070 0.034 ** 0.044 ** 0.005 0.017 0.149 ** 0.202 ** 0.126 − 0.015 0.043 ** 0.045 ** 0.019 0.045 (0.047) (0.060) (0.133) (0.110) (0.017) (0.021) (0.050) (0.043) (0.040) (0.048) (0.104) (0.106) (0.016) (0.019) (0.048) (0.043) ln(1 + EPO Patent Stock )t− 1 0.305 ** 0.283 ** 0.103 0.489 ** 0.025 0.031 0.000 0.059 0.241 ** 0.205 ** 0.043 0.474 ** 0.022 0.037 − 0.009 0.039 (0.052) (0.062) (0.157) (0.123) (0.020) (0.024) (0.056) (0.046) (0.053) (0.062) (0.143) (0.133) (0.019) (0.023) (0.050) (0.045) ln( Net assets ) 0.539 ** 0.535 ** 0.536 ** 0.544 ** 0.613 ** 0.603 ** 0.551 ** 0.661 ** 0.738 ** 0.534 ** 0.533 ** 0.543 ** 0.612 ** 0.602 ** 0.550 ** 0.660 ** (0.012) (0.016) (0.033) (0.021) (0.018) (0.027) (0.034) (0.031) (0.012) (0.016) (0.032) (0.021) (0.018) (0.027) (0.034) (0.031) Two-digit industry dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Country target dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Acquisition year dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes R 2 0.497 0.521 0.519 0.521 0.433 0.409 0.507 0.525 0.498 0.521 0.520 0.522 0.434 0.409 0.507 0.525 Observations 6,562 3,940 1,058 1,564 6,561 3,619 1,134 1,808 6,562 3,940 1,058 1,564 6,561 3,619 1,134 1,808 Notes : This table reports the results of OLS regressions that examine the relationship between patents and firm value for different periods. We distinguish between firms with low and high revenues. Robust standard errors are in brackets. *Significant at 5%. ** Significant at 1%.
1506 S.Belenzon,A.Patacconi/ResearchPolicy42 (2013) 1496–1510
firms’acquisitionvalue,while stronginthefirstsampleperiod,
subsequentlydisappears.
6.2. Target-acquirersimilarity
Thissectionexplorestheextenttowhichourfindingsaredriven
bysimilaritiesbetweenacquiringandtargetfirms.Wefocusontwo
dimensionsofsimilarity:whethertheacquiringandtargetfirms
operateinthesametwo-digitSICcode(productmarket
similar-ity),andwhethertheacquiringandtargetfirmsoperateinthesame
country(geographicproximity).Weconjecturethat,iftheEPOlabel
providesastrongersignalofqualitythantheUSPTOlabel,thenEPO
patentsshouldbeparticularlyvaluable,relativetoUSPTOpatents,
inbetween-industryandbetween-countrydeals.Inthesedeals,in
fact,asymmetricinformationislikelytobeparticularlyprominent,
asacquiringfirmsmaylackspecificindustryknowledgeor
knowl-edgeaboutlocalconditions.Thus,thequalitycertificationprovided
bytheEPOlabelmaybeespeciallyvaluable(Akerlof,1970).
Tables 7and 8 focusonproduct marketsimilarity and
geo-graphicproximity, respectively. Table7 segregates within- and
between-industryacquisitionsusingthetwo-digitSICcodesofthe
acquiringand targetfirms.We findthatover timeEPOpatents
becomemoreimportant,relativetoUSPTOpatents,for
between-industryacquisitionsrelativetowithin-industryacquisitions.This
istrueregardlessofwhetherpatentsareweighed,althoughmuch
strongerUSPTOeffects areobtainedwhenpatentsareweighed.
Table8segregatesdomesticandcross-borderacquisitions.Here,
USPTOpatentsmatteronlyforvalueindomesticdealsandthe
ear-lierperiod.Bycontrast,EPOpatentshaveastrongeffectonvalue
bothin domesticand cross-borderacquisitions.TheriseinEPO
patentvalues,however,isparticularlypronouncedincross-border
acquisitions.
Overall,ourfindings indicatethatEPO patentshavebecome
especiallyimportantinsituationswhereasymmetricinformation
ismostlikelytobesubstantial(i.e.,between-industryand
cross-countrydeals).ThissuggeststhepossibilitythattheEPOlabelmay
providesomeformofqualityreassurance.Exploringthisissue
fur-therisaninterestingdirectionforfutureresearch.
6.3. Targetfirmnationality
Animportantconcern isthat ourtemporal patternsmaybe
drivenbychangesinsamplecomposition.Mostimportantly,the
proportionofEuropeantargetsinoursamplerisessubstantially
overtime.17IfUSPTOpatentsarelessimportantforEuropeanfirms
becausetheymostlysellinEuropeanmarkets,thenapatternof
decliningUSPTOpatentvaluesmightbeobserved.Tomitigatethis
concern,Table9splitsthesamplebytargetfirmnationality(inthis
tablepatentsarealwaysweighedbycitations).Wefindthat,for
Americantargets,USPTOpatentshavealargeeffectonvalueinthe
earliersampleperiod,butthiseffectcompletelydisappearsbythe
endofthesampleperiod.EPOpatents,ontheotherhand,
experi-enceanoppositepattern:theyhavenoeffectonvalueintheearlier
sampleperiod,andalargeeffectinthelaterperiod.ForEuropean
targets,thepatternissimilar.USPTOpatentsareonlyimportant
forvalueintheearliersampleperiods,whileEPOpatentshavea
strongandsignificanteffectinbothearlyandlateperiods(with
muchstrongerlate-periodeffects).Thus,ourpatterns–andin
par-ticularthepatternofdecliningUSPTOeffects–donotseemtobe
17 Thereisastrong“home-bias”effectinacquisitions.84percentofacquisitions byAmericanfirmsareAmerican,andalmostallacquisitionsbyEuropeanfirmsare
European.Thesepatternsarestableovertime. Table
7 Within-vs. between-industry acquisitions. Dependent variable: ln( Firm value ) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) Patent count (not weighed by citations) Patents weighed by citations Target and acquirer are in the same 2-digit SIC code Target and acquirer are in different 2-digit SIC codes Target and acquirer are in the same 2-digit SIC code Target and acquirer are in different 2-digit SIC codes All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 ln(1 + USPTO Patent Stock )t− 1 0.077 ** 0.080 ** 0.062 0.048 0.044 ** 0.051 ** 0.007 − 0.009 0.097 ** 0.100 ** 0.087 0.067 0.064 ** 0.072 ** 0.033 − 0.010 (0.019) (0.022) (0.060) (0.045) (0.015) (0.018) (0.045) (0.037) (0.018) (0.021) (0.052) (0.043) (0.015) (0.018) (0.045) (0.037) ln(1 + EPO Patent Stock )t− 1 0.071 ** 0.084 ** 0.052 0.105 * 0.118 ** 0.105 ** 0.160 ** 0.163 ** 0.054 ** 0.063 ** 0.033 0.092 0.107 ** 0.090 ** 0.152 ** 0.169 ** (0.022) (0.024) (0.069) (0.051) (0.018) (0.021) (0.050) (0.045) (0.021) (0.024) (0.061) (0.048) (0.019) (0.022) (0.048) (0.046) ln( Net assets ) 0.728 ** 0.741 ** 0.711 ** 0.709 ** 0.701 ** 0.704 ** 0.716 ** 0.683 ** 0.727 ** 0.741 ** 0.711 ** 0.709 ** 0.701 ** 0.703 ** 0.716 ** 0.683 ** (0.007) (0.009) (0.015) (0.012) (0.006) (0.008) (0.015) (0.010) (0.006) (0.009) (0.015) (0.012) (0.006) (0.008) (0.015) (0.010) Two-digit industry dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Country target dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Acquisition year dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes R 2 0.700 0.719 0.693 0.694 0.667 0.687 0.664 0.655 0.700 0.720 0.693 0.694 0.667 0.688 0.664 0.656 Observations 13,751 7,805 2,453 3,493 16,555 9,497 2,437 4,621 13,751 7,805 2,453 3,493 16,555 9,497 2,437 4,621 Notes : This table reports the results of OLS regressions that examine the relationship between patents and firm value for different periods. We distinguish between and between-industry acquisitions. Robust standard errors are in brackets. *Significant at 5%. ** Significant at 1%.
S. Belenzon, A. Patacconi / Research Policy 42 (2013) 1496 – 1510 1507
Domestic-vs.cross-borderacquisitions.
Dependentvariable:ln(Firmvalue)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)
Patentcount(notweighedbycitations) Patentsweighedbycitations
Domesticacquisitions Cross-borderacquisitions Domesticacquisitions Cross-borderacquisitions
All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007
ln(1+USPTOPatentStock)t−1 0.066** 0.067** 0.045 0.022 0.006 0.015 −0.014 −0.041 0.086** 0.089** 0.066 0.034 0.035 0.043 0.033 −0.048
(0.013) (0.015) (0.039) (0.032) (0.033) (0.044) (0.101) (0.068) (0.012) (0.014) (0.037) (0.031) (0.035) (0.049) (0.082) (0.070) ln(1+EPOPatentStock)t−1 0.093** 0.092** 0.114** 0.131** 0.155** 0.136** 0.114 0.212** 0.080** 0.074** 0.105** 0.127** 0.134** 0.114* 0.073 0.217** (0.015) (0.017) (0.045) (0.037) (0.036) (0.047) (0.119) (0.076) (0.015) (0.017) (0.042) (0.037) (0.038) (0.051) (0.105) (0.074) ln(Netassets) 0.718** 0.726** 0.709** 0.703** 0.655** 0.659** 0.660** 0.637** 0.717** 0.726** 0.709** 0.703** 0.655** 0.659** 0.659** 0.637** (0.005) (0.006) (0.011) (0.008) (0.014) (0.022) (0.030) (0.025) (0.005) (0.006) (0.011) (0.008) (0.014) (0.022) (0.030) (0.025)
Two-digitindustrydummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Countrytargetdummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Acquisitionyeardummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
R2 0.691 0.710 0.677 0.680 0.607 0.619 0.672 0.615 0.691 0.710 0.677 0.680 0.607 0.619 0.672 0.615 Observations 26,559 15,536 4,092 6,931 3,747 1,766 798 1,183 26,559 15,536 4,092 6,931 3,747 1,766 798 1,183 Notes:ThistablereportstheresultsofOLSregressionsthatexaminetheeffectofpatentsonfirmvaluefordomesticvs.cross-borderacquisitions.Robuststandarderrorsareinbrackets.
*Significantat5%.
**Significantat1%.
Table9
American-vs.Europeantargets.
Dependentvariable:ln(Firmvalue)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16)
Patentcount(notweighedbycitations) Patentsweighedbycitations
Americantarget Europeantarget Americantarget Europeantarget
All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007 All 1985–1998 1999–2001 2002–2007
ln(1+USPTOPatentStock)t−1 0.041** 0.045** 0.007 0.031 0.056** 0.069** 0.070 −0.102 0.061** 0.071** 0.023 0.036 0.083** 0.095** 0.118* −0.123*
(0.016) (0.018) (0.054) (0.047) (0.021) (0.024) (0.064) (0.058) (0.015) (0.017) (0.045) (0.042) (0.021) (0.024) (0.059) (0.062) ln(1+EPOPatentStock)t−1 0.105** 0.105** 0.145** 0.095* 0.101** 0.096** 0.064 0.216** 0.090** 0.084** 0.136** 0.096* 0.085** 0.077** 0.039 0.227** (0.018) (0.020) (0.064) (0.055) (0.025) (0.029) (0.068) (0.059) (0.018) (0.020) (0.058) (0.054) (0.026) (0.029) (0.065) (0.061) ln(Netassets) 0.719** 0.744** 0.648** 0.684** 0.707** 0.699** 0.741** 0.700** 0.718** 0.742** 0.648** 0.684** 0.707** 0.699** 0.741** 0.700** (0.007) (0.008) (0.020) (0.017) (0.006) (0.009) (0.013) (0.010) (0.007) (0.008) (0.020) (0.017) (0.006) (0.009) (0.013) (0.010)
Two-digitindustrydummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Countrytargetdummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Acquisitionyeardummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
R2 0.675 0.713 0.563 0.652 0.694 0.697 0.739 0.688 0.676 0.713 0.563 0.652 0.694 0.697 0.740 0.688 Observations 12,001 8,170 1,866 1,965 14,655 8,067 2,476 4,112 12,001 8,170 1,866 1,965 14,655 8,067 2,476 4,112 Notes:ThistablereportstheresultsofOLSregressionsthatexaminetheeffectofpatentsonfirmvalue.WesplitthesampledependingonwhethertargetsareAmericanorEuropeanfirms.Robuststandarderrorsareinbrackets.
*Significantat5%.